Date: Wed, 18 Mar 1998 23:41:26 -0500 (EST) From: "John S. Dyson" <toor@dyson.iquest.net> To: karl@mcs.net (Karl Denninger) Cc: dannyman@sasquatch.dannyland.org, dyson@FreeBSD.ORG, current@FreeBSD.ORG Subject: Re: A reminder of toxic -current Message-ID: <199803190441.XAA00612@dyson.iquest.net> In-Reply-To: <19980318210037.16941@mcs.net> from Karl Denninger at "Mar 18, 98 09:00:37 pm"
next in thread | previous in thread | raw e-mail | index | archive | help
> On Wed, Mar 18, 1998 at 08:47:24PM -0600, dannyman wrote: > > On Wed, Mar 18, 1998 at 09:16:56PM -0500, John S. Dyson wrote: > > > Warning, todays -current is still toxic to your filesystems. I > > > am still working on it, trying to isolate the problem(s). > > > > ermmm, i recently rebuilt a second time after the FS scare, and neither of my > > -current's have shown any trouble. i've backed up my home directory just in > > case, but if my fs hasn't been destroyed yet, and is stable enough to reliably > > recompile the OS, I should worry? > > > > -dan > > YES! > > I can reproduce this failure within 30 minutes, and it DESTROYS the > filesystem in question. FSCK will report it clean after a couple of passes, > but the data contained in it is irrecoverably damaged. > > DO NOT USE CURRENT KERNELS BUILT AFTER THE MEGACOMMIT FROM JOHN OVER THIS > LAST WEEKEND! > Karl is right. This will be the last warning from me regarding the -current problem. I am not going to broadcast any proclaimations of a corrected kernel but I am working on fixing the problems, with the help of Karl and others. If Karl and others find that the problems are fixed, the information will be slowly exposed. This is one reason that it is said that -current is not for the faint of heart, and assuming that it will work in a production environment without full testing is very dangerous. I didn't even know that my changes had problems, until my system became unusable. That ended up delaying my warnings for over 24 hours, and people found out the hard-way. The bugs are my fault, but please don't assume that because -current normally works OK, that it is always safe to use in production. It is best to know when to cut your losses, and it is best just to back up to a 12-mar or before kernel. Not doing that will simply put your filesystems at risk... So, one-more-time, if you aren't working with Karl or me (or whomever else is trying to solve the problem), please do not use a kernel after 12-mar. I apologize for the inconvieniences, but fretting instead of simply backing up to a 12-mar kernel isn't productive. Karl knows how to protect his systems, and other people who try to use -current in production know how to do it also. Even then, it is possible to get totally hosed, like I did... If people are running the -current kernel without having local copies of the CVS tree and/or backup copies of various kernels, well, that is like playing with fire. FYI, I have probably 100-200 kernel trees that I keep online, and perhaps that is overkill, but keeping your last 10-20 safe kernels isn't a terrible idea. John To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199803190441.XAA00612>